Linear methods in statistics, as applied to biomedical data analysis, continue to be studied. The study of optimal linear model estimates for variance component estimation continues to be an area of investigation. An extensive study of repeated-measures experiments was begun, and the results will be a monograph on the unified exposition of the problem as well as to present an exact multivariate alternative analysis. Further, linear, nonparametric, multivariate methods have also been outlined. Additionally, a new approach to the problem of describing and characterizing multivariate dependence structures has been presented that ties the practical problem of studying dependence to a geometric framework.